US retail media is closing in on $70B. Every retailer you buy from is quietly an ad network.
US retail media ad spend hit $58.79B last year and eMarketer has it at $69.33B this year — up nearly 18%, against about 8% growth for US digital advertising overall.
Which means the grocery chain, the pharmacy and the big-box store you shop at are all, structurally, media companies now. The retail is the audience-acquisition strategy.
Tinuiti's Q4 numbers are worth sitting with. Amazon Sponsored Products clicks up 23% year over year while cost-per-click actually fell 1%. Amazon DSP spend up 31% and now 40% of total Amazon ad budgets. Walmart Sponsored Products up 14%, with 60% of its self-serve display spend going offsite.
That last one is the tell. Retail media isn't staying on the retailer's site anymore. It's following you around the internet using the retailer's purchase data.
The case FOR leaning in:
Closed-loop measurement. You can see the sale. After a decade of attribution arguments, that's a relief
You're reaching people at the moment of intent, not the moment of scroll
Offsite retail media gives you purchase-data targeting that third-party cookies never actually delivered
The case AGAINST:
You're paying the retailer for access to customers you already brought them. A lot of retail media spend is defensive — buying your own brand term so a competitor doesn't
Incrementality is genuinely murky. Closed-loop attribution proves correlation with a sale, not that you caused it
It's a margin transfer. Every dollar of retail media profit came out of a brand's P&L, and retailers know exactly how much leverage they have
Fragmentation. Dozens of networks, no common standard, and a reporting burden that eats your team
Where I land: retail media is now a cost of distribution, not a marketing channel — and you should budget it that way.
Stop comparing it to paid social ROAS and start comparing it to trade spend and slotting fees, because that's what it functionally replaced. Then fight hard on incrementality testing, because that's the only lever you have.
The brands doing this well run holdout tests and are ruthless about switching things off. The rest are paying rent.
Anyone running proper incrementality tests on retail media — what happened when you switched it off? I've heard some very uncomfortable answers privately.
Notes on the data
One correction to something circulating widely, including in my own decks until recently: 2025 retail media spend was $58.79B, not "over $60B." Small thing, but the $60B version is everywhere and it's wrong.
Also note the Tinuiti growth figures are their own managed portfolio — $4B+ of spend, so a real sample, but not industry-wide.